Along with the rapid advance in technology, the digital content industry, which includes computer animation, digital game, digital learning, mobile application and service, also has gained rapid growth. The visual based man-machine interface system mainly captures images with camera, and further detects the user's movement or gesture through image recognition to interact with the digital content. However, the conventional single-camera interactive system may easily end up with low recognition rate due to color closeness or background interference.
Following the conventional single-camera interactive system, the dual-camera man-machine interface system is disclosed. The dual-camera man-machine interface system calculates the left and right parallax to create a depth image, not only effectively capturing the object belonging to the foreground with respect to the camera but also filtering the background interference. However, a large computation volume is required for creating the depth image, and real-time computation cannot be provided for ordinary personal computer under high resolution requirement (such as QVGA 320×240 resolution). Therefore, the dual-camera interactive system must limit the resolution level to comply with the requirement of the real-time computation. However, with the limited resolution level, the interactive system can only recognize the user's large limbs and segments (such as head and hand), but cannot provide more delicate interactive operation.